OpenAI began testing advertising inside ChatGPT on Monday, introducing paid placements for users on its Free and Go plans in the United States as the company looks for new ways to fund the fast growing costs of building and running large scale AI systems.
The ads will appear only for people using ChatGPT for free and for subscribers on the Go tier, a lower cost plan launched globally in January that costs $8 a month in the U.S. OpenAI said users on its higher priced subscriptions, including Plus, Pro, Business, Enterprise, and Education, will not see advertising.
The company said the test is limited to the U.S. and that it is still evaluating how ads fit into a conversational AI product. OpenAI did not say how many users are included in the trial or how long it will last.
OpenAI framed the move as a way to expand access while protecting trust. In a blog post, the company said ads will not affect how ChatGPT answers questions and that advertisers will not see individual user conversations. It said ads will be clearly labeled as sponsored and kept separate from the chatbot’s regular responses.
The decision comes as OpenAI faces heavy financial pressure from the cost of training and running large AI models. Operating ChatGPT at global scale requires vast amounts of computing power, energy, and specialized hardware. While OpenAI earns money from subscriptions and enterprise deals, those revenues have not fully offset expenses tied to rapid expansion.
ChatGPT now has hundreds of millions of users worldwide, making it one of the most widely used consumer AI products ever launched. Turning part of that audience into an advertising base mirrors the business models of search engines and social media platforms, which use ads to support free services.
OpenAI said ads in ChatGPT will be shown based on context and past interactions. For example, someone asking about cooking might see promotions related to food or grocery delivery. The company said advertisers will only receive aggregated performance data, such as how many people saw or clicked an ad, and not personal user information.
Users will be able to dismiss ads, give feedback, and manage ad personalization settings. OpenAI also said users can view their ad interaction history and clear it at any time. Ads will not be shown to users under 18 and will not appear near topics the company classifies as sensitive or regulated, such as health, politics, or mental health.
OpenAI is entering a space where user trust is central. Unlike traditional websites or social media feeds, ChatGPT is used for direct problem solving, writing, coding, research, and personal tasks. Many users rely on it for guidance that feels private and conversational. That makes the placement of ads more complex than simply adding banners to a webpage.
Last year, OpenAI tested in app suggestions that some users felt crossed the line into advertising. The company later scaled those back after complaints. This time, OpenAI says it is taking a more formal and transparent approach, with clear labeling and user controls.
The rollout also highlights how competitive the AI market has become. OpenAI faces growing pressure from rivals like Anthropic, Google, Meta, and several open source model builders. Anthropic, which makes the Claude chatbot, aired Super Bowl commercials on Sunday that mocked the idea of ads inside AI tools, showing fictional chatbots awkwardly mixing advice with irrelevant promotions.
OpenAI Chief Executive Sam Altman criticized those ads, saying they misrepresented how OpenAI plans to use advertising. Anthropic has not publicly responded to those remarks.
Behind the public exchanges is a larger struggle over how AI companies pay for infrastructure at scale. Training and running models like GPT-4 and its successors costs billions of dollars over time. Microsoft, OpenAI’s main investor and partner, has poured large sums into the company and provides cloud infrastructure through Azure. Microsoft itself earns much of its revenue from enterprise software and advertising through Bing and other services.
Read also: OpenAI: 7.5% VAT On ChatGPT Plus Services, Begins In Nigeria
If ads in ChatGPT expand beyond testing, they could place OpenAI in more direct competition with Google and Meta in the digital advertising market. Unlike search ads, which appear alongside lists of links, ChatGPT ads appear in a conversational environment where users expect focused, personalized answers.
That raises questions about how people will respond to commercial messages inside what feels like a private dialogue. OpenAI says it will not allow advertisers to shape or influence responses. It says the AI’s output remains independent of any ad relationship.
OpenAI also said it is not selling user data to advertisers. Instead, ads are matched to general topics and past interactions within ChatGPT itself. The company said it does not allow advertisers to target people based on identity or outside data sources.
The Go plan, which now includes ads, is OpenAI’s first attempt at a low cost paid tier. It sits between the free version and the $20 a month Plus plan. By adding ads to Go and Free users, OpenAI is testing whether it can offer lower prices while still generating revenue.
So far, OpenAI has not said whether ads will eventually appear outside the U.S. or whether they will become permanent. The company said it will study user feedback, engagement, and the overall effect on trust before making broader decisions.
The introduction of ads marks a shift in how ChatGPT is positioned. Until now, the product has mostly avoided overt commercial elements. The test suggests OpenAI is moving toward a model closer to other large consumer platforms, where free access is supported by advertising and premium access removes it.
OpenAI said it will update its policies as the test continues and that user input will shape what happens next. For now, the company is positioning the ad rollout as an experiment rather than a final design.
The next formal step will be evaluation. OpenAI said it will review how users interact with ads, how often they are dismissed, and whether they interfere with how people use ChatGPT. The company said it will decide on any wider rollout only after that process is complete.
